1 /* $NetBSD: nd6_rtr.c,v 1.77 2008/12/19 18:49:39 cegger Exp $ */
2 /* $KAME: nd6_rtr.c,v 1.95 2001/02/07 08:09:47 itojun Exp $ */
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 #include <sys/cdefs.h>
34 __KERNEL_RCSID(0, "$NetBSD: nd6_rtr.c,v 1.77 2008/12/19 18:49:39 cegger Exp $");
36 #include <sys/param.h>
37 #include <sys/systm.h>
38 #include <sys/malloc.h>
40 #include <sys/socket.h>
41 #include <sys/sockio.h>
43 #include <sys/kernel.h>
44 #include <sys/errno.h>
45 #include <sys/ioctl.h>
46 #include <sys/syslog.h>
49 #include <net/if_types.h>
50 #include <net/if_dl.h>
51 #include <net/route.h>
52 #include <net/radix.h>
54 #include <netinet/in.h>
55 #include <netinet6/in6_var.h>
56 #include <netinet6/in6_ifattach.h>
57 #include <netinet/ip6.h>
58 #include <netinet6/ip6_var.h>
59 #include <netinet6/nd6.h>
60 #include <netinet/icmp6.h>
61 #include <netinet6/icmp6_private.h>
62 #include <netinet6/scope6_var.h>
64 #include <net/net_osdep.h>
66 static int rtpref(struct nd_defrouter
*);
67 static struct nd_defrouter
*defrtrlist_update(struct nd_defrouter
*);
68 static int prelist_update(struct nd_prefixctl
*, struct nd_defrouter
*,
70 static struct in6_ifaddr
*in6_ifadd(struct nd_prefixctl
*, int);
71 static struct nd_pfxrouter
*pfxrtr_lookup(struct nd_prefix
*,
72 struct nd_defrouter
*);
73 static void pfxrtr_add(struct nd_prefix
*, struct nd_defrouter
*);
74 static void pfxrtr_del(struct nd_pfxrouter
*);
75 static struct nd_pfxrouter
*find_pfxlist_reachable_router
77 static void defrouter_delreq(struct nd_defrouter
*);
78 static void nd6_rtmsg(int, struct rtentry
*);
80 static int in6_init_prefix_ltimes(struct nd_prefix
*);
81 static void in6_init_address_ltimes(struct nd_prefix
*ndpr
,
82 struct in6_addrlifetime
*lt6
);
84 static int rt6_deleteroute(struct rtentry
*, void *);
86 extern int nd6_recalc_reachtm_interval
;
88 static struct ifnet
*nd6_defifp
;
91 int ip6_use_tempaddr
= 0;
93 int ip6_desync_factor
;
94 u_int32_t ip6_temp_preferred_lifetime
= DEF_TEMP_PREFERRED_LIFETIME
;
95 u_int32_t ip6_temp_valid_lifetime
= DEF_TEMP_VALID_LIFETIME
;
96 int ip6_temp_regen_advance
= TEMPADDR_REGEN_ADVANCE
;
98 /* RTPREF_MEDIUM has to be 0! */
100 #define RTPREF_MEDIUM 0
101 #define RTPREF_LOW (-1)
102 #define RTPREF_RESERVED (-2)
103 #define RTPREF_INVALID (-3) /* internal */
106 * Receive Router Solicitation Message - just for routers.
107 * Router solicitation/advertisement is mostly managed by userland program
108 * (rtadvd) so here we have no function like nd6_ra_output().
113 nd6_rs_input(struct mbuf
*m
, int off
, int icmp6len
)
115 struct ifnet
*ifp
= m
->m_pkthdr
.rcvif
;
116 struct ip6_hdr
*ip6
= mtod(m
, struct ip6_hdr
*);
117 struct nd_router_solicit
*nd_rs
;
118 struct in6_addr saddr6
= ip6
->ip6_src
;
121 union nd_opts ndopts
;
123 /* If I'm not a router, ignore it. */
124 if (ip6_accept_rtadv
!= 0 || !ip6_forwarding
)
128 if (ip6
->ip6_hlim
!= 255) {
130 "nd6_rs_input: invalid hlim (%d) from %s to %s on %s\n",
131 ip6
->ip6_hlim
, ip6_sprintf(&ip6
->ip6_src
),
132 ip6_sprintf(&ip6
->ip6_dst
), if_name(ifp
)));
137 * Don't update the neighbor cache, if src = ::.
138 * This indicates that the src has no IP address assigned yet.
140 if (IN6_IS_ADDR_UNSPECIFIED(&saddr6
))
143 IP6_EXTHDR_GET(nd_rs
, struct nd_router_solicit
*, m
, off
, icmp6len
);
145 ICMP6_STATINC(ICMP6_STAT_TOOSHORT
);
149 icmp6len
-= sizeof(*nd_rs
);
150 nd6_option_init(nd_rs
+ 1, icmp6len
, &ndopts
);
151 if (nd6_options(&ndopts
) < 0) {
153 "nd6_rs_input: invalid ND option, ignored\n"));
154 /* nd6_options have incremented stats */
158 if (ndopts
.nd_opts_src_lladdr
) {
159 lladdr
= (char *)(ndopts
.nd_opts_src_lladdr
+ 1);
160 lladdrlen
= ndopts
.nd_opts_src_lladdr
->nd_opt_len
<< 3;
163 if (lladdr
&& ((ifp
->if_addrlen
+ 2 + 7) & ~7) != lladdrlen
) {
165 "nd6_rs_input: lladdrlen mismatch for %s "
166 "(if %d, RS packet %d)\n",
167 ip6_sprintf(&saddr6
), ifp
->if_addrlen
, lladdrlen
- 2));
171 nd6_cache_lladdr(ifp
, &saddr6
, lladdr
, lladdrlen
, ND_ROUTER_SOLICIT
, 0);
178 ICMP6_STATINC(ICMP6_STAT_BADRS
);
183 * Receive Router Advertisement Message.
186 * TODO: on-link bit on prefix information
187 * TODO: ND_RA_FLAG_{OTHER,MANAGED} processing
190 nd6_ra_input(struct mbuf
*m
, int off
, int icmp6len
)
192 struct ifnet
*ifp
= m
->m_pkthdr
.rcvif
;
193 struct nd_ifinfo
*ndi
= ND_IFINFO(ifp
);
194 struct ip6_hdr
*ip6
= mtod(m
, struct ip6_hdr
*);
195 struct nd_router_advert
*nd_ra
;
196 struct in6_addr saddr6
= ip6
->ip6_src
;
198 struct in6_addr daddr6
= ip6
->ip6_dst
;
199 int flags
; /* = nd_ra->nd_ra_flags_reserved; */
200 int is_managed
= ((flags
& ND_RA_FLAG_MANAGED
) != 0);
201 int is_other
= ((flags
& ND_RA_FLAG_OTHER
) != 0);
204 union nd_opts ndopts
;
205 struct nd_defrouter
*dr
;
208 * We only accept RAs only when
209 * the system-wide variable allows the acceptance, and
210 * per-interface variable allows RAs on the receiving interface.
212 if (ip6_accept_rtadv
== 0)
214 if (!(ndi
->flags
& ND6_IFF_ACCEPT_RTADV
))
217 if (ip6
->ip6_hlim
!= 255) {
219 "nd6_ra_input: invalid hlim (%d) from %s to %s on %s\n",
220 ip6
->ip6_hlim
, ip6_sprintf(&ip6
->ip6_src
),
221 ip6_sprintf(&ip6
->ip6_dst
), if_name(ifp
)));
225 if (!IN6_IS_ADDR_LINKLOCAL(&saddr6
)) {
227 "nd6_ra_input: src %s is not link-local\n",
228 ip6_sprintf(&saddr6
)));
232 IP6_EXTHDR_GET(nd_ra
, struct nd_router_advert
*, m
, off
, icmp6len
);
234 ICMP6_STATINC(ICMP6_STAT_TOOSHORT
);
238 icmp6len
-= sizeof(*nd_ra
);
239 nd6_option_init(nd_ra
+ 1, icmp6len
, &ndopts
);
240 if (nd6_options(&ndopts
) < 0) {
242 "nd6_ra_input: invalid ND option, ignored\n"));
243 /* nd6_options have incremented stats */
248 struct nd_defrouter drtr
;
249 u_int32_t advreachable
= nd_ra
->nd_ra_reachable
;
251 /* remember if this is a multicasted advertisement */
252 if (IN6_IS_ADDR_MULTICAST(&ip6
->ip6_dst
))
255 memset(&drtr
, 0, sizeof(drtr
));
256 drtr
.rtaddr
= saddr6
;
257 drtr
.flags
= nd_ra
->nd_ra_flags_reserved
;
258 drtr
.rtlifetime
= ntohs(nd_ra
->nd_ra_router_lifetime
);
259 drtr
.expire
= time_second
+ drtr
.rtlifetime
;
261 /* unspecified or not? (RFC 2461 6.3.4) */
264 if (advreachable
<= MAX_REACHABLE_TIME
&&
265 ndi
->basereachable
!= advreachable
) {
266 ndi
->basereachable
= advreachable
;
267 ndi
->reachable
= ND_COMPUTE_RTIME(ndi
->basereachable
);
268 ndi
->recalctm
= nd6_recalc_reachtm_interval
; /* reset */
271 if (nd_ra
->nd_ra_retransmit
)
272 ndi
->retrans
= ntohl(nd_ra
->nd_ra_retransmit
);
273 if (nd_ra
->nd_ra_curhoplimit
)
274 ndi
->chlim
= nd_ra
->nd_ra_curhoplimit
;
275 dr
= defrtrlist_update(&drtr
);
281 if (ndopts
.nd_opts_pi
) {
282 struct nd_opt_hdr
*pt
;
283 struct nd_opt_prefix_info
*pi
= NULL
;
284 struct nd_prefixctl pr
;
286 for (pt
= (struct nd_opt_hdr
*)ndopts
.nd_opts_pi
;
287 pt
<= (struct nd_opt_hdr
*)ndopts
.nd_opts_pi_end
;
288 pt
= (struct nd_opt_hdr
*)((char *)pt
+
289 (pt
->nd_opt_len
<< 3))) {
290 if (pt
->nd_opt_type
!= ND_OPT_PREFIX_INFORMATION
)
292 pi
= (struct nd_opt_prefix_info
*)pt
;
294 if (pi
->nd_opt_pi_len
!= 4) {
296 "nd6_ra_input: invalid option "
297 "len %d for prefix information option, "
298 "ignored\n", pi
->nd_opt_pi_len
));
302 if (128 < pi
->nd_opt_pi_prefix_len
) {
304 "nd6_ra_input: invalid prefix "
305 "len %d for prefix information option, "
306 "ignored\n", pi
->nd_opt_pi_prefix_len
));
310 if (IN6_IS_ADDR_MULTICAST(&pi
->nd_opt_pi_prefix
)
311 || IN6_IS_ADDR_LINKLOCAL(&pi
->nd_opt_pi_prefix
)) {
313 "nd6_ra_input: invalid prefix "
315 ip6_sprintf(&pi
->nd_opt_pi_prefix
)));
319 memset(&pr
, 0, sizeof(pr
));
320 sockaddr_in6_init(&pr
.ndpr_prefix
,
321 &pi
->nd_opt_pi_prefix
, 0, 0, 0);
322 pr
.ndpr_ifp
= (struct ifnet
*)m
->m_pkthdr
.rcvif
;
324 pr
.ndpr_raf_onlink
= (pi
->nd_opt_pi_flags_reserved
&
325 ND_OPT_PI_FLAG_ONLINK
) ? 1 : 0;
326 pr
.ndpr_raf_auto
= (pi
->nd_opt_pi_flags_reserved
&
327 ND_OPT_PI_FLAG_AUTO
) ? 1 : 0;
328 pr
.ndpr_plen
= pi
->nd_opt_pi_prefix_len
;
329 pr
.ndpr_vltime
= ntohl(pi
->nd_opt_pi_valid_time
);
330 pr
.ndpr_pltime
= ntohl(pi
->nd_opt_pi_preferred_time
);
332 (void)prelist_update(&pr
, dr
, m
, mcast
);
339 if (ndopts
.nd_opts_mtu
&& ndopts
.nd_opts_mtu
->nd_opt_mtu_len
== 1) {
343 mtu
= ntohl(ndopts
.nd_opts_mtu
->nd_opt_mtu_mtu
);
346 if (mtu
< IPV6_MMTU
) {
347 nd6log((LOG_INFO
, "nd6_ra_input: bogus mtu option "
348 "mtu=%lu sent from %s, ignoring\n",
349 mtu
, ip6_sprintf(&ip6
->ip6_src
)));
354 maxmtu
= (ndi
->maxmtu
&& ndi
->maxmtu
< ifp
->if_mtu
)
355 ? ndi
->maxmtu
: ifp
->if_mtu
;
357 int change
= (ndi
->linkmtu
!= mtu
);
360 if (change
) /* in6_maxmtu may change */
363 nd6log((LOG_INFO
, "nd6_ra_input: bogus mtu "
364 "mtu=%lu sent from %s; "
365 "exceeds maxmtu %lu, ignoring\n",
366 mtu
, ip6_sprintf(&ip6
->ip6_src
), maxmtu
));
373 * Source link layer address
379 if (ndopts
.nd_opts_src_lladdr
) {
380 lladdr
= (char *)(ndopts
.nd_opts_src_lladdr
+ 1);
381 lladdrlen
= ndopts
.nd_opts_src_lladdr
->nd_opt_len
<< 3;
384 if (lladdr
&& ((ifp
->if_addrlen
+ 2 + 7) & ~7) != lladdrlen
) {
386 "nd6_ra_input: lladdrlen mismatch for %s "
387 "(if %d, RA packet %d)\n", ip6_sprintf(&saddr6
),
388 ifp
->if_addrlen
, lladdrlen
- 2));
392 nd6_cache_lladdr(ifp
, &saddr6
, lladdr
, lladdrlen
, ND_ROUTER_ADVERT
, 0);
395 * Installing a link-layer address might change the state of the
396 * router's neighbor cache, which might also affect our on-link
397 * detection of adveritsed prefixes.
399 pfxlist_onlink_check();
407 ICMP6_STATINC(ICMP6_STAT_BADRA
);
412 * default router list processing sub routines
415 /* tell the change to user processes watching the routing socket. */
417 nd6_rtmsg(int cmd
, struct rtentry
*rt
)
419 struct rt_addrinfo info
;
421 memset((void *)&info
, 0, sizeof(info
));
422 info
.rti_info
[RTAX_DST
] = rt_getkey(rt
);
423 info
.rti_info
[RTAX_GATEWAY
] = rt
->rt_gateway
;
424 info
.rti_info
[RTAX_NETMASK
] = rt_mask(rt
);
426 info
.rti_info
[RTAX_IFP
] = rt
->rt_ifp
->if_dl
->ifa_addr
;
427 info
.rti_info
[RTAX_IFA
] = rt
->rt_ifa
->ifa_addr
;
430 rt_missmsg(cmd
, &info
, rt
->rt_flags
, 0);
434 defrouter_addreq(struct nd_defrouter
*new)
437 struct sockaddr_in6 sin6
;
440 struct rtentry
*newrt
= NULL
;
444 memset(&def
, 0, sizeof(def
));
445 memset(&mask
, 0, sizeof(mask
));
446 memset(&gate
, 0,sizeof(gate
)); /* for safety */
448 def
.sin6
.sin6_len
= mask
.sin6
.sin6_len
= gate
.sin6
.sin6_len
=
449 sizeof(struct sockaddr_in6
);
450 def
.sin6
.sin6_family
= mask
.sin6
.sin6_family
= gate
.sin6
.sin6_family
= AF_INET6
;
451 gate
.sin6
.sin6_addr
= new->rtaddr
;
452 #ifndef SCOPEDROUTING
453 gate
.sin6
.sin6_scope_id
= 0; /* XXX */
457 error
= rtrequest(RTM_ADD
, &def
.sa
, &gate
.sa
, &mask
.sa
,
458 RTF_GATEWAY
, &newrt
);
460 nd6_rtmsg(RTM_ADD
, newrt
); /* tell user process */
469 struct nd_defrouter
*
470 defrouter_lookup(const struct in6_addr
*addr
, struct ifnet
*ifp
)
472 struct nd_defrouter
*dr
;
474 TAILQ_FOREACH(dr
, &nd_defrouter
, dr_entry
) {
475 if (dr
->ifp
== ifp
&& IN6_ARE_ADDR_EQUAL(addr
, &dr
->rtaddr
))
479 return dr
; /* search failed */
483 defrtrlist_del(struct nd_defrouter
*dr
)
485 struct nd_defrouter
*deldr
= NULL
;
486 struct nd_prefix
*pr
;
489 * Flush all the routing table entries that use the router
492 if (!ip6_forwarding
&& ip6_accept_rtadv
) /* XXX: better condition? */
493 rt6_flush(&dr
->rtaddr
, dr
->ifp
);
497 defrouter_delreq(dr
);
499 TAILQ_REMOVE(&nd_defrouter
, dr
, dr_entry
);
502 * Also delete all the pointers to the router in each prefix lists.
504 LIST_FOREACH(pr
, &nd_prefix
, ndpr_entry
) {
505 struct nd_pfxrouter
*pfxrtr
;
506 if ((pfxrtr
= pfxrtr_lookup(pr
, dr
)) != NULL
)
509 pfxlist_onlink_check();
512 * If the router is the primary one, choose a new one.
513 * Note that defrouter_select() will remove the current gateway
514 * from the routing table.
523 * Remove the default route for a given router.
524 * This is just a subroutine function for defrouter_select(), and should
525 * not be called from anywhere else.
528 defrouter_delreq(struct nd_defrouter
*dr
)
531 struct sockaddr_in6 sin6
;
534 struct rtentry
*oldrt
= NULL
;
538 panic("dr == NULL in defrouter_delreq");
541 memset(&def
, 0, sizeof(def
));
542 memset(&mask
, 0, sizeof(mask
));
543 memset(&gw
, 0, sizeof(gw
)); /* for safety */
545 def
.sin6
.sin6_len
= mask
.sin6
.sin6_len
= gw
.sin6
.sin6_len
=
546 sizeof(struct sockaddr_in6
);
547 def
.sin6
.sin6_family
= mask
.sin6
.sin6_family
= gw
.sin6
.sin6_family
= AF_INET6
;
548 gw
.sin6
.sin6_addr
= dr
->rtaddr
;
549 #ifndef SCOPEDROUTING
550 gw
.sin6
.sin6_scope_id
= 0; /* XXX */
553 rtrequest(RTM_DELETE
, &def
.sa
, &gw
.sa
, &mask
.sa
, RTF_GATEWAY
, &oldrt
);
555 nd6_rtmsg(RTM_DELETE
, oldrt
);
556 if (oldrt
->rt_refcnt
<= 0) {
558 * XXX: borrowed from the RTM_DELETE case of
570 * remove all default routes from default router list
573 defrouter_reset(void)
575 struct nd_defrouter
*dr
;
577 for (dr
= TAILQ_FIRST(&nd_defrouter
); dr
;
578 dr
= TAILQ_NEXT(dr
, dr_entry
))
579 defrouter_delreq(dr
);
582 * XXX should we also nuke any default routers in the kernel, by
583 * going through them by rtalloc1()?
588 * Default Router Selection according to Section 6.3.6 of RFC 2461 and
589 * draft-ietf-ipngwg-router-selection:
590 * 1) Routers that are reachable or probably reachable should be preferred.
591 * If we have more than one (probably) reachable router, prefer ones
592 * with the highest router preference.
593 * 2) When no routers on the list are known to be reachable or
594 * probably reachable, routers SHOULD be selected in a round-robin
595 * fashion, regardless of router preference values.
596 * 3) If the Default Router List is empty, assume that all
597 * destinations are on-link.
599 * We assume nd_defrouter is sorted by router preference value.
600 * Since the code below covers both with and without router preference cases,
601 * we do not need to classify the cases by ifdef.
603 * At this moment, we do not try to install more than one default router,
604 * even when the multipath routing is available, because we're not sure about
605 * the benefits for stub hosts comparing to the risk of making the code
606 * complicated and the possibility of introducing bugs.
609 defrouter_select(void)
611 int s
= splsoftnet();
612 struct nd_defrouter
*dr
, *selected_dr
= NULL
, *installed_dr
= NULL
;
613 struct rtentry
*rt
= NULL
;
614 struct llinfo_nd6
*ln
= NULL
;
617 * This function should be called only when acting as an autoconfigured
618 * host. Although the remaining part of this function is not effective
619 * if the node is not an autoconfigured host, we explicitly exclude
620 * such cases here for safety.
622 if (ip6_forwarding
|| !ip6_accept_rtadv
) {
624 "defrouter_select: called unexpectedly (forwarding=%d, "
625 "accept_rtadv=%d)\n", ip6_forwarding
, ip6_accept_rtadv
));
631 * Let's handle easy case (3) first:
632 * If default router list is empty, there's nothing to be done.
634 if (!TAILQ_FIRST(&nd_defrouter
)) {
640 * Search for a (probably) reachable router from the list.
641 * We just pick up the first reachable one (if any), assuming that
642 * the ordering rule of the list described in defrtrlist_update().
644 for (dr
= TAILQ_FIRST(&nd_defrouter
); dr
;
645 dr
= TAILQ_NEXT(dr
, dr_entry
)) {
646 if (selected_dr
== NULL
&&
647 (rt
= nd6_lookup(&dr
->rtaddr
, 0, dr
->ifp
)) != NULL
&&
648 (ln
= (struct llinfo_nd6
*)rt
->rt_llinfo
) != NULL
&&
649 ND6_IS_LLINFO_PROBREACH(ln
)) {
653 if (dr
->installed
&& !installed_dr
)
655 else if (dr
->installed
&& installed_dr
) {
656 /* this should not happen. warn for diagnosis. */
657 log(LOG_ERR
, "defrouter_select: more than one router"
662 * If none of the default routers was found to be reachable,
663 * round-robin the list regardless of preference.
664 * Otherwise, if we have an installed router, check if the selected
665 * (reachable) router should really be preferred to the installed one.
666 * We only prefer the new router when the old one is not reachable
667 * or when the new one has a really higher preference value.
669 if (selected_dr
== NULL
) {
670 if (installed_dr
== NULL
|| !TAILQ_NEXT(installed_dr
, dr_entry
))
671 selected_dr
= TAILQ_FIRST(&nd_defrouter
);
673 selected_dr
= TAILQ_NEXT(installed_dr
, dr_entry
);
674 } else if (installed_dr
&&
675 (rt
= nd6_lookup(&installed_dr
->rtaddr
, 0, installed_dr
->ifp
)) &&
676 (ln
= (struct llinfo_nd6
*)rt
->rt_llinfo
) &&
677 ND6_IS_LLINFO_PROBREACH(ln
) &&
678 rtpref(selected_dr
) <= rtpref(installed_dr
)) {
679 selected_dr
= installed_dr
;
683 * If the selected router is different than the installed one,
684 * remove the installed router and install the selected one.
685 * Note that the selected router is never NULL here.
687 if (installed_dr
!= selected_dr
) {
689 defrouter_delreq(installed_dr
);
690 defrouter_addreq(selected_dr
);
698 * for default router selection
699 * regards router-preference field as a 2-bit signed integer
702 rtpref(struct nd_defrouter
*dr
)
704 switch (dr
->flags
& ND_RA_FLAG_RTPREF_MASK
) {
705 case ND_RA_FLAG_RTPREF_HIGH
:
706 return (RTPREF_HIGH
);
707 case ND_RA_FLAG_RTPREF_MEDIUM
:
708 case ND_RA_FLAG_RTPREF_RSV
:
709 return (RTPREF_MEDIUM
);
710 case ND_RA_FLAG_RTPREF_LOW
:
714 * This case should never happen. If it did, it would mean a
715 * serious bug of kernel internal. We thus always bark here.
716 * Or, can we even panic?
718 log(LOG_ERR
, "rtpref: impossible RA flag %x\n", dr
->flags
);
719 return (RTPREF_INVALID
);
724 static struct nd_defrouter
*
725 defrtrlist_update(struct nd_defrouter
*new)
727 struct nd_defrouter
*dr
, *n
;
728 int s
= splsoftnet();
730 if ((dr
= defrouter_lookup(&new->rtaddr
, new->ifp
)) != NULL
) {
732 if (new->rtlifetime
== 0) {
736 int oldpref
= rtpref(dr
);
739 dr
->flags
= new->flags
; /* xxx flag check */
740 dr
->rtlifetime
= new->rtlifetime
;
741 dr
->expire
= new->expire
;
744 * If the preference does not change, there's no need
745 * to sort the entries.
747 if (rtpref(new) == oldpref
) {
753 * preferred router may be changed, so relocate
755 * XXX: calling TAILQ_REMOVE directly is a bad manner.
756 * However, since defrtrlist_del() has many side
757 * effects, we intentionally do so here.
758 * defrouter_select() below will handle routing
761 TAILQ_REMOVE(&nd_defrouter
, dr
, dr_entry
);
769 /* entry does not exist */
770 if (new->rtlifetime
== 0) {
775 n
= (struct nd_defrouter
*)malloc(sizeof(*n
), M_IP6NDP
, M_NOWAIT
);
780 memset(n
, 0, sizeof(*n
));
785 * Insert the new router in the Default Router List;
786 * The Default Router List should be in the descending order
787 * of router-preferece. Routers with the same preference are
788 * sorted in the arriving time order.
791 /* insert at the end of the group */
792 for (dr
= TAILQ_FIRST(&nd_defrouter
); dr
;
793 dr
= TAILQ_NEXT(dr
, dr_entry
)) {
794 if (rtpref(n
) > rtpref(dr
))
798 TAILQ_INSERT_BEFORE(dr
, n
, dr_entry
);
800 TAILQ_INSERT_TAIL(&nd_defrouter
, n
, dr_entry
);
809 static struct nd_pfxrouter
*
810 pfxrtr_lookup(struct nd_prefix
*pr
, struct nd_defrouter
*dr
)
812 struct nd_pfxrouter
*search
;
814 LIST_FOREACH(search
, &pr
->ndpr_advrtrs
, pfr_entry
) {
815 if (search
->router
== dr
)
823 pfxrtr_add(struct nd_prefix
*pr
, struct nd_defrouter
*dr
)
825 struct nd_pfxrouter
*new;
827 new = malloc(sizeof(*new), M_IP6NDP
, M_NOWAIT
|M_ZERO
);
832 LIST_INSERT_HEAD(&pr
->ndpr_advrtrs
, new, pfr_entry
);
834 pfxlist_onlink_check();
838 pfxrtr_del(struct nd_pfxrouter
*pfr
)
840 LIST_REMOVE(pfr
, pfr_entry
);
845 nd6_prefix_lookup(struct nd_prefixctl
*key
)
847 struct nd_prefix
*search
;
849 LIST_FOREACH(search
, &nd_prefix
, ndpr_entry
) {
850 if (key
->ndpr_ifp
== search
->ndpr_ifp
&&
851 key
->ndpr_plen
== search
->ndpr_plen
&&
852 in6_are_prefix_equal(&key
->ndpr_prefix
.sin6_addr
,
853 &search
->ndpr_prefix
.sin6_addr
, key
->ndpr_plen
)) {
862 nd6_prelist_add(struct nd_prefixctl
*pr
, struct nd_defrouter
*dr
,
863 struct nd_prefix
**newp
)
865 struct nd_prefix
*new = NULL
;
870 new = malloc(sizeof(*new), M_IP6NDP
, M_NOWAIT
|M_ZERO
);
873 new->ndpr_ifp
= pr
->ndpr_ifp
;
874 new->ndpr_prefix
= pr
->ndpr_prefix
;
875 new->ndpr_plen
= pr
->ndpr_plen
;
876 new->ndpr_vltime
= pr
->ndpr_vltime
;
877 new->ndpr_pltime
= pr
->ndpr_pltime
;
878 new->ndpr_flags
= pr
->ndpr_flags
;
879 if ((error
= in6_init_prefix_ltimes(new)) != 0) {
883 new->ndpr_lastupdate
= time_second
;
888 LIST_INIT(&new->ndpr_advrtrs
);
889 in6_prefixlen2mask(&new->ndpr_mask
, new->ndpr_plen
);
890 /* make prefix in the canonical form */
891 for (i
= 0; i
< 4; i
++)
892 new->ndpr_prefix
.sin6_addr
.s6_addr32
[i
] &=
893 new->ndpr_mask
.s6_addr32
[i
];
896 /* link ndpr_entry to nd_prefix list */
897 LIST_INSERT_HEAD(&nd_prefix
, new, ndpr_entry
);
900 /* ND_OPT_PI_FLAG_ONLINK processing */
901 if (new->ndpr_raf_onlink
) {
904 if ((e
= nd6_prefix_onlink(new)) != 0) {
905 nd6log((LOG_ERR
, "nd6_prelist_add: failed to make "
906 "the prefix %s/%d on-link on %s (errno=%d)\n",
907 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
908 pr
->ndpr_plen
, if_name(pr
->ndpr_ifp
), e
));
909 /* proceed anyway. XXX: is it correct? */
920 prelist_remove(struct nd_prefix
*pr
)
922 struct nd_pfxrouter
*pfr
, *next
;
925 /* make sure to invalidate the prefix until it is really freed. */
930 * Though these flags are now meaningless, we'd rather keep the value
931 * not to confuse users when executing "ndp -p".
933 pr
->ndpr_raf_onlink
= 0;
934 pr
->ndpr_raf_auto
= 0;
936 if ((pr
->ndpr_stateflags
& NDPRF_ONLINK
) != 0 &&
937 (e
= nd6_prefix_offlink(pr
)) != 0) {
938 nd6log((LOG_ERR
, "prelist_remove: failed to make %s/%d offlink "
940 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
941 pr
->ndpr_plen
, if_name(pr
->ndpr_ifp
), e
));
942 /* what should we do? */
945 if (pr
->ndpr_refcnt
> 0)
946 return; /* notice here? */
949 /* unlink ndpr_entry from nd_prefix list */
950 LIST_REMOVE(pr
, ndpr_entry
);
952 /* free list of routers that adversed the prefix */
953 for (pfr
= LIST_FIRST(&pr
->ndpr_advrtrs
); pfr
!= NULL
; pfr
= next
) {
954 next
= LIST_NEXT(pfr
, pfr_entry
);
962 pfxlist_onlink_check();
966 prelist_update(struct nd_prefixctl
*new,
967 struct nd_defrouter
*dr
, /* may be NULL */
971 struct in6_ifaddr
*ia6
= NULL
, *ia6_match
= NULL
;
973 struct ifnet
*ifp
= new->ndpr_ifp
;
974 struct nd_prefix
*pr
;
975 int s
= splsoftnet();
979 struct in6_addrlifetime lt6_tmp
;
984 * Authenticity for NA consists authentication for
985 * both IP header and IP datagrams, doesn't it ?
987 #if defined(M_AUTHIPHDR) && defined(M_AUTHIPDGM)
988 auth
= (m
->m_flags
& M_AUTHIPHDR
989 && m
->m_flags
& M_AUTHIPDGM
) ? 1 : 0;
993 if ((pr
= nd6_prefix_lookup(new)) != NULL
) {
995 * nd6_prefix_lookup() ensures that pr and new have the same
996 * prefix on a same interface.
1000 * Update prefix information. Note that the on-link (L) bit
1001 * and the autonomous (A) bit should NOT be changed from 1
1004 if (new->ndpr_raf_onlink
== 1)
1005 pr
->ndpr_raf_onlink
= 1;
1006 if (new->ndpr_raf_auto
== 1)
1007 pr
->ndpr_raf_auto
= 1;
1008 if (new->ndpr_raf_onlink
) {
1009 pr
->ndpr_vltime
= new->ndpr_vltime
;
1010 pr
->ndpr_pltime
= new->ndpr_pltime
;
1011 (void)in6_init_prefix_ltimes(pr
); /* XXX error case? */
1012 pr
->ndpr_lastupdate
= time_second
;
1015 if (new->ndpr_raf_onlink
&&
1016 (pr
->ndpr_stateflags
& NDPRF_ONLINK
) == 0) {
1019 if ((e
= nd6_prefix_onlink(pr
)) != 0) {
1021 "prelist_update: failed to make "
1022 "the prefix %s/%d on-link on %s "
1024 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
1025 pr
->ndpr_plen
, if_name(pr
->ndpr_ifp
), e
));
1026 /* proceed anyway. XXX: is it correct? */
1030 if (dr
&& pfxrtr_lookup(pr
, dr
) == NULL
)
1033 struct nd_prefix
*newpr
= NULL
;
1037 if (new->ndpr_vltime
== 0)
1039 if (new->ndpr_raf_onlink
== 0 && new->ndpr_raf_auto
== 0)
1042 error
= nd6_prelist_add(new, dr
, &newpr
);
1043 if (error
!= 0 || newpr
== NULL
) {
1044 nd6log((LOG_NOTICE
, "prelist_update: "
1045 "nd6_prelist_add failed for %s/%d on %s "
1046 "errno=%d, returnpr=%p\n",
1047 ip6_sprintf(&new->ndpr_prefix
.sin6_addr
),
1048 new->ndpr_plen
, if_name(new->ndpr_ifp
),
1050 goto end
; /* we should just give up in this case. */
1054 * XXX: from the ND point of view, we can ignore a prefix
1055 * with the on-link bit being zero. However, we need a
1056 * prefix structure for references from autoconfigured
1057 * addresses. Thus, we explicitly make sure that the prefix
1058 * itself expires now.
1060 if (newpr
->ndpr_raf_onlink
== 0) {
1061 newpr
->ndpr_vltime
= 0;
1062 newpr
->ndpr_pltime
= 0;
1063 in6_init_prefix_ltimes(newpr
);
1070 * Address autoconfiguration based on Section 5.5.3 of RFC 2462.
1071 * Note that pr must be non NULL at this point.
1074 /* 5.5.3 (a). Ignore the prefix without the A bit set. */
1075 if (!new->ndpr_raf_auto
)
1079 * 5.5.3 (b). the link-local prefix should have been ignored in
1083 /* 5.5.3 (c). Consistency check on lifetimes: pltime <= vltime. */
1084 if (new->ndpr_pltime
> new->ndpr_vltime
) {
1085 error
= EINVAL
; /* XXX: won't be used */
1090 * 5.5.3 (d). If the prefix advertised is not equal to the prefix of
1091 * an address configured by stateless autoconfiguration already in the
1092 * list of addresses associated with the interface, and the Valid
1093 * Lifetime is not 0, form an address. We first check if we have
1094 * a matching prefix.
1095 * Note: we apply a clarification in rfc2462bis-02 here. We only
1096 * consider autoconfigured addresses while RFC2462 simply said
1099 IFADDR_FOREACH(ifa
, ifp
) {
1100 struct in6_ifaddr
*ifa6
;
1101 u_int32_t remaininglifetime
;
1103 if (ifa
->ifa_addr
->sa_family
!= AF_INET6
)
1106 ifa6
= (struct in6_ifaddr
*)ifa
;
1109 * We only consider autoconfigured addresses as per rfc2462bis.
1111 if (!(ifa6
->ia6_flags
& IN6_IFF_AUTOCONF
))
1115 * Spec is not clear here, but I believe we should concentrate
1116 * on unicast (i.e. not anycast) addresses.
1117 * XXX: other ia6_flags? detached or duplicated?
1119 if ((ifa6
->ia6_flags
& IN6_IFF_ANYCAST
) != 0)
1123 * Ignore the address if it is not associated with a prefix
1124 * or is associated with a prefix that is different from this
1125 * one. (pr is never NULL here)
1127 if (ifa6
->ia6_ndpr
!= pr
)
1130 if (ia6_match
== NULL
) /* remember the first one */
1134 * An already autoconfigured address matched. Now that we
1135 * are sure there is at least one matched address, we can
1136 * proceed to 5.5.3. (e): update the lifetimes according to the
1137 * "two hours" rule and the privacy extension.
1138 * We apply some clarifications in rfc2462bis:
1139 * - use remaininglifetime instead of storedlifetime as a
1141 * - remove the dead code in the "two-hour" rule
1143 #define TWOHOUR (120*60)
1144 lt6_tmp
= ifa6
->ia6_lifetime
;
1145 if (lt6_tmp
.ia6t_vltime
== ND6_INFINITE_LIFETIME
)
1146 remaininglifetime
= ND6_INFINITE_LIFETIME
;
1147 else if (time_second
- ifa6
->ia6_updatetime
>
1148 lt6_tmp
.ia6t_vltime
) {
1150 * The case of "invalid" address. We should usually
1151 * not see this case.
1153 remaininglifetime
= 0;
1155 remaininglifetime
= lt6_tmp
.ia6t_vltime
-
1156 (time_second
- ifa6
->ia6_updatetime
);
1158 /* when not updating, keep the current stored lifetime. */
1159 lt6_tmp
.ia6t_vltime
= remaininglifetime
;
1161 if (TWOHOUR
< new->ndpr_vltime
||
1162 remaininglifetime
< new->ndpr_vltime
) {
1163 lt6_tmp
.ia6t_vltime
= new->ndpr_vltime
;
1164 } else if (remaininglifetime
<= TWOHOUR
) {
1166 lt6_tmp
.ia6t_vltime
= new->ndpr_vltime
;
1169 * new->ndpr_vltime <= TWOHOUR &&
1170 * TWOHOUR < remaininglifetime
1172 lt6_tmp
.ia6t_vltime
= TWOHOUR
;
1175 /* The 2 hour rule is not imposed for preferred lifetime. */
1176 lt6_tmp
.ia6t_pltime
= new->ndpr_pltime
;
1178 in6_init_address_ltimes(pr
, <6_tmp
);
1181 * We need to treat lifetimes for temporary addresses
1182 * differently, according to
1183 * draft-ietf-ipv6-privacy-addrs-v2-01.txt 3.3 (1);
1184 * we only update the lifetimes when they are in the maximum
1187 if ((ifa6
->ia6_flags
& IN6_IFF_TEMPORARY
) != 0) {
1188 u_int32_t maxvltime
, maxpltime
;
1190 if (ip6_temp_valid_lifetime
>
1191 (u_int32_t
)((time_second
- ifa6
->ia6_createtime
) +
1192 ip6_desync_factor
)) {
1193 maxvltime
= ip6_temp_valid_lifetime
-
1194 (time_second
- ifa6
->ia6_createtime
) -
1198 if (ip6_temp_preferred_lifetime
>
1199 (u_int32_t
)((time_second
- ifa6
->ia6_createtime
) +
1200 ip6_desync_factor
)) {
1201 maxpltime
= ip6_temp_preferred_lifetime
-
1202 (time_second
- ifa6
->ia6_createtime
) -
1207 if (lt6_tmp
.ia6t_vltime
== ND6_INFINITE_LIFETIME
||
1208 lt6_tmp
.ia6t_vltime
> maxvltime
) {
1209 lt6_tmp
.ia6t_vltime
= maxvltime
;
1211 if (lt6_tmp
.ia6t_pltime
== ND6_INFINITE_LIFETIME
||
1212 lt6_tmp
.ia6t_pltime
> maxpltime
) {
1213 lt6_tmp
.ia6t_pltime
= maxpltime
;
1217 ifa6
->ia6_lifetime
= lt6_tmp
;
1218 ifa6
->ia6_updatetime
= time_second
;
1220 if (ia6_match
== NULL
&& new->ndpr_vltime
) {
1224 * 5.5.3 (d) (continued)
1225 * No address matched and the valid lifetime is non-zero.
1226 * Create a new address.
1230 * Prefix Length check:
1231 * If the sum of the prefix length and interface identifier
1232 * length does not equal 128 bits, the Prefix Information
1233 * option MUST be ignored. The length of the interface
1234 * identifier is defined in a separate link-type specific
1237 ifidlen
= in6_if2idlen(ifp
);
1239 /* this should not happen, so we always log it. */
1240 log(LOG_ERR
, "prelist_update: IFID undefined (%s)\n",
1244 if (ifidlen
+ pr
->ndpr_plen
!= 128) {
1246 "prelist_update: invalid prefixlen "
1247 "%d for %s, ignored\n",
1248 pr
->ndpr_plen
, if_name(ifp
)));
1252 if ((ia6
= in6_ifadd(new, mcast
)) != NULL
) {
1254 * note that we should use pr (not new) for reference.
1260 * draft-ietf-ipngwg-temp-addresses-v2-00 3.3 (2).
1261 * When a new public address is created as described
1262 * in RFC2462, also create a new temporary address.
1264 * draft-ietf-ipngwg-temp-addresses-v2-00 3.5.
1265 * When an interface connects to a new link, a new
1266 * randomized interface identifier should be generated
1267 * immediately together with a new set of temporary
1268 * addresses. Thus, we specifiy 1 as the 2nd arg of
1271 if (ip6_use_tempaddr
) {
1273 if ((e
= in6_tmpifadd(ia6
, 1, 1)) != 0) {
1274 nd6log((LOG_NOTICE
, "prelist_update: "
1275 "failed to create a temporary "
1276 "address, errno=%d\n",
1282 * A newly added address might affect the status
1283 * of other addresses, so we check and update it.
1284 * XXX: what if address duplication happens?
1286 pfxlist_onlink_check();
1288 /* just set an error. do not bark here. */
1289 error
= EADDRNOTAVAIL
; /* XXX: might be unused. */
1299 * A supplement function used in the on-link detection below;
1300 * detect if a given prefix has a (probably) reachable advertising router.
1301 * XXX: lengthy function name...
1303 static struct nd_pfxrouter
*
1304 find_pfxlist_reachable_router(struct nd_prefix
*pr
)
1306 struct nd_pfxrouter
*pfxrtr
;
1308 struct llinfo_nd6
*ln
;
1310 for (pfxrtr
= LIST_FIRST(&pr
->ndpr_advrtrs
); pfxrtr
;
1311 pfxrtr
= LIST_NEXT(pfxrtr
, pfr_entry
)) {
1312 if ((rt
= nd6_lookup(&pfxrtr
->router
->rtaddr
, 0,
1313 pfxrtr
->router
->ifp
)) &&
1314 (ln
= (struct llinfo_nd6
*)rt
->rt_llinfo
) &&
1315 ND6_IS_LLINFO_PROBREACH(ln
))
1323 * Check if each prefix in the prefix list has at least one available router
1324 * that advertised the prefix (a router is "available" if its neighbor cache
1325 * entry is reachable or probably reachable).
1326 * If the check fails, the prefix may be off-link, because, for example,
1327 * we have moved from the network but the lifetime of the prefix has not
1328 * expired yet. So we should not use the prefix if there is another prefix
1329 * that has an available router.
1330 * But, if there is no prefix that has an available router, we still regards
1331 * all the prefixes as on-link. This is because we can't tell if all the
1332 * routers are simply dead or if we really moved from the network and there
1333 * is no router around us.
1336 pfxlist_onlink_check(void)
1338 struct nd_prefix
*pr
;
1339 struct in6_ifaddr
*ifa
;
1340 struct nd_defrouter
*dr
;
1341 struct nd_pfxrouter
*pfxrtr
= NULL
;
1344 * Check if there is a prefix that has a reachable advertising
1347 LIST_FOREACH(pr
, &nd_prefix
, ndpr_entry
) {
1348 if (pr
->ndpr_raf_onlink
&& find_pfxlist_reachable_router(pr
))
1352 * If we have no such prefix, check whether we still have a router
1353 * that does not advertise any prefixes.
1356 TAILQ_FOREACH(dr
, &nd_defrouter
, dr_entry
) {
1357 struct nd_prefix
*pr0
;
1359 LIST_FOREACH(pr0
, &nd_prefix
, ndpr_entry
) {
1360 if ((pfxrtr
= pfxrtr_lookup(pr0
, dr
)) != NULL
)
1367 if (pr
!= NULL
|| (TAILQ_FIRST(&nd_defrouter
) && !pfxrtr
)) {
1369 * There is at least one prefix that has a reachable router,
1370 * or at least a router which probably does not advertise
1371 * any prefixes. The latter would be the case when we move
1372 * to a new link where we have a router that does not provide
1373 * prefixes and we configure an address by hand.
1374 * Detach prefixes which have no reachable advertising
1375 * router, and attach other prefixes.
1377 LIST_FOREACH(pr
, &nd_prefix
, ndpr_entry
) {
1378 /* XXX: a link-local prefix should never be detached */
1379 if (IN6_IS_ADDR_LINKLOCAL(&pr
->ndpr_prefix
.sin6_addr
))
1383 * we aren't interested in prefixes without the L bit
1386 if (pr
->ndpr_raf_onlink
== 0)
1389 if ((pr
->ndpr_stateflags
& NDPRF_DETACHED
) == 0 &&
1390 find_pfxlist_reachable_router(pr
) == NULL
)
1391 pr
->ndpr_stateflags
|= NDPRF_DETACHED
;
1392 if ((pr
->ndpr_stateflags
& NDPRF_DETACHED
) != 0 &&
1393 find_pfxlist_reachable_router(pr
) != 0)
1394 pr
->ndpr_stateflags
&= ~NDPRF_DETACHED
;
1397 /* there is no prefix that has a reachable router */
1398 LIST_FOREACH(pr
, &nd_prefix
, ndpr_entry
) {
1399 if (IN6_IS_ADDR_LINKLOCAL(&pr
->ndpr_prefix
.sin6_addr
))
1402 if (pr
->ndpr_raf_onlink
== 0)
1405 if ((pr
->ndpr_stateflags
& NDPRF_DETACHED
) != 0)
1406 pr
->ndpr_stateflags
&= ~NDPRF_DETACHED
;
1411 * Remove each interface route associated with a (just) detached
1412 * prefix, and reinstall the interface route for a (just) attached
1413 * prefix. Note that all attempt of reinstallation does not
1414 * necessarily success, when a same prefix is shared among multiple
1415 * interfaces. Such cases will be handled in nd6_prefix_onlink,
1416 * so we don't have to care about them.
1418 LIST_FOREACH(pr
, &nd_prefix
, ndpr_entry
) {
1421 if (IN6_IS_ADDR_LINKLOCAL(&pr
->ndpr_prefix
.sin6_addr
))
1424 if (pr
->ndpr_raf_onlink
== 0)
1427 if ((pr
->ndpr_stateflags
& NDPRF_DETACHED
) != 0 &&
1428 (pr
->ndpr_stateflags
& NDPRF_ONLINK
) != 0) {
1429 if ((e
= nd6_prefix_offlink(pr
)) != 0) {
1431 "pfxlist_onlink_check: failed to "
1432 "make %s/%d offlink, errno=%d\n",
1433 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
1437 if ((pr
->ndpr_stateflags
& NDPRF_DETACHED
) == 0 &&
1438 (pr
->ndpr_stateflags
& NDPRF_ONLINK
) == 0 &&
1439 pr
->ndpr_raf_onlink
) {
1440 if ((e
= nd6_prefix_onlink(pr
)) != 0) {
1442 "pfxlist_onlink_check: failed to "
1443 "make %s/%d onlink, errno=%d\n",
1444 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
1451 * Changes on the prefix status might affect address status as well.
1452 * Make sure that all addresses derived from an attached prefix are
1453 * attached, and that all addresses derived from a detached prefix are
1454 * detached. Note, however, that a manually configured address should
1455 * always be attached.
1456 * The precise detection logic is same as the one for prefixes.
1458 for (ifa
= in6_ifaddr
; ifa
; ifa
= ifa
->ia_next
) {
1459 if (!(ifa
->ia6_flags
& IN6_IFF_AUTOCONF
))
1462 if (ifa
->ia6_ndpr
== NULL
) {
1464 * This can happen when we first configure the address
1465 * (i.e. the address exists, but the prefix does not).
1466 * XXX: complicated relationships...
1471 if (find_pfxlist_reachable_router(ifa
->ia6_ndpr
))
1475 for (ifa
= in6_ifaddr
; ifa
; ifa
= ifa
->ia_next
) {
1476 if ((ifa
->ia6_flags
& IN6_IFF_AUTOCONF
) == 0)
1479 if (ifa
->ia6_ndpr
== NULL
) /* XXX: see above. */
1482 if (find_pfxlist_reachable_router(ifa
->ia6_ndpr
)) {
1483 if (ifa
->ia6_flags
& IN6_IFF_DETACHED
) {
1484 ifa
->ia6_flags
&= ~IN6_IFF_DETACHED
;
1485 ifa
->ia6_flags
|= IN6_IFF_TENTATIVE
;
1486 nd6_dad_start((struct ifaddr
*)ifa
,
1490 if ((ifa
->ia6_flags
& IN6_IFF_DETACHED
) == 0) {
1491 ifa
->ia6_flags
|= IN6_IFF_DETACHED
;
1497 for (ifa
= in6_ifaddr
; ifa
; ifa
= ifa
->ia_next
) {
1498 if ((ifa
->ia6_flags
& IN6_IFF_AUTOCONF
) == 0)
1501 if (ifa
->ia6_flags
& IN6_IFF_DETACHED
) {
1502 ifa
->ia6_flags
&= ~IN6_IFF_DETACHED
;
1503 ifa
->ia6_flags
|= IN6_IFF_TENTATIVE
;
1504 /* Do we need a delay in this case? */
1505 nd6_dad_start((struct ifaddr
*)ifa
, 0);
1512 nd6_prefix_onlink(struct nd_prefix
*pr
)
1515 struct ifnet
*ifp
= pr
->ndpr_ifp
;
1516 struct sockaddr_in6 mask6
;
1517 struct nd_prefix
*opr
;
1520 struct rtentry
*rt
= NULL
;
1523 if ((pr
->ndpr_stateflags
& NDPRF_ONLINK
) != 0) {
1525 "nd6_prefix_onlink: %s/%d is already on-link\n",
1526 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
), pr
->ndpr_plen
));
1531 * Add the interface route associated with the prefix. Before
1532 * installing the route, check if there's the same prefix on another
1533 * interface, and the prefix has already installed the interface route.
1534 * Although such a configuration is expected to be rare, we explicitly
1537 LIST_FOREACH(opr
, &nd_prefix
, ndpr_entry
) {
1541 if ((opr
->ndpr_stateflags
& NDPRF_ONLINK
) == 0)
1544 if (opr
->ndpr_plen
== pr
->ndpr_plen
&&
1545 in6_are_prefix_equal(&pr
->ndpr_prefix
.sin6_addr
,
1546 &opr
->ndpr_prefix
.sin6_addr
, pr
->ndpr_plen
))
1551 * We prefer link-local addresses as the associated interface address.
1553 /* search for a link-local addr */
1554 ifa
= (struct ifaddr
*)in6ifa_ifpforlinklocal(ifp
,
1555 IN6_IFF_NOTREADY
| IN6_IFF_ANYCAST
);
1557 /* XXX: freebsd does not have ifa_ifwithaf */
1558 IFADDR_FOREACH(ifa
, ifp
) {
1559 if (ifa
->ifa_addr
->sa_family
== AF_INET6
)
1562 /* should we care about ia6_flags? */
1566 * This can still happen, when, for example, we receive an RA
1567 * containing a prefix with the L bit set and the A bit clear,
1568 * after removing all IPv6 addresses on the receiving
1569 * interface. This should, of course, be rare though.
1572 "nd6_prefix_onlink: failed to find any ifaddr"
1573 " to add route for a prefix(%s/%d) on %s\n",
1574 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
1575 pr
->ndpr_plen
, if_name(ifp
)));
1580 * in6_ifinit() sets nd6_rtrequest to ifa_rtrequest for all ifaddrs.
1581 * ifa->ifa_rtrequest = nd6_rtrequest;
1583 memset(&mask6
, 0, sizeof(mask6
));
1584 mask6
.sin6_len
= sizeof(mask6
);
1585 mask6
.sin6_addr
= pr
->ndpr_mask
;
1586 /* rtrequest() will probably set RTF_UP, but we're not sure. */
1587 rtflags
= ifa
->ifa_flags
| RTF_UP
;
1588 if (nd6_need_cache(ifp
)) {
1589 /* explicitly set in case ifa_flags does not set the flag. */
1590 rtflags
|= RTF_CLONING
;
1593 * explicitly clear the cloning bit in case ifa_flags sets it.
1595 rtflags
&= ~RTF_CLONING
;
1597 error
= rtrequest(RTM_ADD
, (struct sockaddr
*)&pr
->ndpr_prefix
,
1598 ifa
->ifa_addr
, (struct sockaddr
*)&mask6
, rtflags
, &rt
);
1600 if (rt
!= NULL
) /* this should be non NULL, though */
1601 nd6_rtmsg(RTM_ADD
, rt
);
1602 pr
->ndpr_stateflags
|= NDPRF_ONLINK
;
1604 nd6log((LOG_ERR
, "nd6_prefix_onlink: failed to add route for a"
1605 " prefix (%s/%d) on %s, gw=%s, mask=%s, flags=%lx "
1607 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
),
1608 pr
->ndpr_plen
, if_name(ifp
),
1609 ip6_sprintf(&((struct sockaddr_in6
*)ifa
->ifa_addr
)->sin6_addr
),
1610 ip6_sprintf(&mask6
.sin6_addr
), rtflags
, error
));
1620 nd6_prefix_offlink(struct nd_prefix
*pr
)
1623 struct ifnet
*ifp
= pr
->ndpr_ifp
;
1624 struct nd_prefix
*opr
;
1625 struct sockaddr_in6 sa6
, mask6
;
1626 struct rtentry
*rt
= NULL
;
1629 if ((pr
->ndpr_stateflags
& NDPRF_ONLINK
) == 0) {
1631 "nd6_prefix_offlink: %s/%d is already off-link\n",
1632 ip6_sprintf(&pr
->ndpr_prefix
.sin6_addr
), pr
->ndpr_plen
));
1636 sockaddr_in6_init(&sa6
, &pr
->ndpr_prefix
.sin6_addr
, 0, 0, 0);
1637 sockaddr_in6_init(&mask6
, &pr
->ndpr_mask
, 0, 0, 0);
1638 error
= rtrequest(RTM_DELETE
, (struct sockaddr
*)&sa6
, NULL
,
1639 (struct sockaddr
*)&mask6
, 0, &rt
);
1641 pr
->ndpr_stateflags
&= ~NDPRF_ONLINK
;
1643 /* report the route deletion to the routing socket. */
1645 nd6_rtmsg(RTM_DELETE
, rt
);
1648 * There might be the same prefix on another interface,
1649 * the prefix which could not be on-link just because we have
1650 * the interface route (see comments in nd6_prefix_onlink).
1651 * If there's one, try to make the prefix on-link on the
1654 LIST_FOREACH(opr
, &nd_prefix
, ndpr_entry
) {
1658 if ((opr
->ndpr_stateflags
& NDPRF_ONLINK
) != 0)
1662 * KAME specific: detached prefixes should not be
1665 if ((opr
->ndpr_stateflags
& NDPRF_DETACHED
) != 0)
1668 if (opr
->ndpr_plen
== pr
->ndpr_plen
&&
1669 in6_are_prefix_equal(&pr
->ndpr_prefix
.sin6_addr
,
1670 &opr
->ndpr_prefix
.sin6_addr
, pr
->ndpr_plen
)) {
1673 if ((e
= nd6_prefix_onlink(opr
)) != 0) {
1675 "nd6_prefix_offlink: failed to "
1676 "recover a prefix %s/%d from %s "
1677 "to %s (errno = %d)\n",
1678 ip6_sprintf(&opr
->ndpr_prefix
.sin6_addr
),
1679 opr
->ndpr_plen
, if_name(ifp
),
1680 if_name(opr
->ndpr_ifp
), e
));
1685 /* XXX: can we still set the NDPRF_ONLINK flag? */
1687 "nd6_prefix_offlink: failed to delete route: "
1688 "%s/%d on %s (errno = %d)\n",
1689 ip6_sprintf(&sa6
.sin6_addr
), pr
->ndpr_plen
, if_name(ifp
),
1694 if (rt
->rt_refcnt
<= 0) {
1695 /* XXX: we should free the entry ourselves. */
1704 static struct in6_ifaddr
*
1705 in6_ifadd(struct nd_prefixctl
*pr
, int mcast
)
1707 struct ifnet
*ifp
= pr
->ndpr_ifp
;
1709 struct in6_aliasreq ifra
;
1710 struct in6_ifaddr
*ia
, *ib
;
1712 struct in6_addr mask
;
1713 int prefixlen
= pr
->ndpr_plen
;
1716 in6_prefixlen2mask(&mask
, prefixlen
);
1719 * find a link-local address (will be interface ID).
1720 * Is it really mandatory? Theoretically, a global or a site-local
1721 * address can be configured without a link-local address, if we
1722 * have a unique interface identifier...
1724 * it is not mandatory to have a link-local address, we can generate
1725 * interface identifier on the fly. we do this because:
1726 * (1) it should be the easiest way to find interface identifier.
1727 * (2) RFC2462 5.4 suggesting the use of the same interface identifier
1728 * for multiple addresses on a single interface, and possible shortcut
1729 * of DAD. we omitted DAD for this reason in the past.
1730 * (3) a user can prevent autoconfiguration of global address
1731 * by removing link-local address by hand (this is partly because we
1732 * don't have other way to control the use of IPv6 on an interface.
1733 * this has been our design choice - cf. NRL's "ifconfig auto").
1734 * (4) it is easier to manage when an interface has addresses
1735 * with the same interface identifier, than to have multiple addresses
1736 * with different interface identifiers.
1738 ifa
= (struct ifaddr
*)in6ifa_ifpforlinklocal(ifp
, 0); /* 0 is OK? */
1740 ib
= (struct in6_ifaddr
*)ifa
;
1744 #if 0 /* don't care link local addr state, and always do DAD */
1745 /* if link-local address is not eligible, do not autoconfigure. */
1746 if (((struct in6_ifaddr
*)ifa
)->ia6_flags
& IN6_IFF_NOTREADY
) {
1747 printf("in6_ifadd: link-local address not ready\n");
1752 /* prefixlen + ifidlen must be equal to 128 */
1753 plen0
= in6_mask2len(&ib
->ia_prefixmask
.sin6_addr
, NULL
);
1754 if (prefixlen
!= plen0
) {
1755 nd6log((LOG_INFO
, "in6_ifadd: wrong prefixlen for %s "
1756 "(prefix=%d ifid=%d)\n",
1757 if_name(ifp
), prefixlen
, 128 - plen0
));
1763 memset(&ifra
, 0, sizeof(ifra
));
1765 * in6_update_ifa() does not use ifra_name, but we accurately set it
1768 strncpy(ifra
.ifra_name
, if_name(ifp
), sizeof(ifra
.ifra_name
));
1769 sockaddr_in6_init(&ifra
.ifra_addr
, &pr
->ndpr_prefix
.sin6_addr
, 0, 0, 0);
1771 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[0] &= mask
.s6_addr32
[0];
1772 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[1] &= mask
.s6_addr32
[1];
1773 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[2] &= mask
.s6_addr32
[2];
1774 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[3] &= mask
.s6_addr32
[3];
1777 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[0] |=
1778 (ib
->ia_addr
.sin6_addr
.s6_addr32
[0] & ~mask
.s6_addr32
[0]);
1779 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[1] |=
1780 (ib
->ia_addr
.sin6_addr
.s6_addr32
[1] & ~mask
.s6_addr32
[1]);
1781 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[2] |=
1782 (ib
->ia_addr
.sin6_addr
.s6_addr32
[2] & ~mask
.s6_addr32
[2]);
1783 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[3] |=
1784 (ib
->ia_addr
.sin6_addr
.s6_addr32
[3] & ~mask
.s6_addr32
[3]);
1786 /* new prefix mask. */
1787 sockaddr_in6_init(&ifra
.ifra_prefixmask
, &mask
, 0, 0, 0);
1790 ifra
.ifra_lifetime
.ia6t_vltime
= pr
->ndpr_vltime
;
1791 ifra
.ifra_lifetime
.ia6t_pltime
= pr
->ndpr_pltime
;
1793 /* XXX: scope zone ID? */
1795 ifra
.ifra_flags
|= IN6_IFF_AUTOCONF
; /* obey autoconf */
1798 * Make sure that we do not have this address already. This should
1799 * usually not happen, but we can still see this case, e.g., if we
1800 * have manually configured the exact address to be configured.
1802 if (in6ifa_ifpwithaddr(ifp
, &ifra
.ifra_addr
.sin6_addr
) != NULL
) {
1803 /* this should be rare enough to make an explicit log */
1804 log(LOG_INFO
, "in6_ifadd: %s is already configured\n",
1805 ip6_sprintf(&ifra
.ifra_addr
.sin6_addr
));
1810 * Allocate ifaddr structure, link into chain, etc.
1811 * If we are going to create a new address upon receiving a multicasted
1812 * RA, we need to impose a random delay before starting DAD.
1813 * [draft-ietf-ipv6-rfc2462bis-02.txt, Section 5.4.2]
1817 updateflags
|= IN6_IFAUPDATE_DADDELAY
;
1818 if ((error
= in6_update_ifa(ifp
, &ifra
, NULL
, updateflags
)) != 0) {
1820 "in6_ifadd: failed to make ifaddr %s on %s (errno=%d)\n",
1821 ip6_sprintf(&ifra
.ifra_addr
.sin6_addr
), if_name(ifp
),
1823 return (NULL
); /* ifaddr must not have been allocated. */
1826 ia
= in6ifa_ifpwithaddr(ifp
, &ifra
.ifra_addr
.sin6_addr
);
1828 return (ia
); /* this is always non-NULL */
1833 const struct in6_ifaddr
*ia0
, /* corresponding public address */
1837 struct ifnet
*ifp
= ia0
->ia_ifa
.ifa_ifp
;
1838 struct in6_ifaddr
*newia
, *ia
;
1839 struct in6_aliasreq ifra
;
1841 int trylimit
= 3; /* XXX: adhoc value */
1843 u_int32_t randid
[2];
1844 u_int32_t vltime0
, pltime0
;
1846 memset(&ifra
, 0, sizeof(ifra
));
1847 strncpy(ifra
.ifra_name
, if_name(ifp
), sizeof(ifra
.ifra_name
));
1848 ifra
.ifra_addr
= ia0
->ia_addr
;
1849 /* copy prefix mask */
1850 ifra
.ifra_prefixmask
= ia0
->ia_prefixmask
;
1851 /* clear the old IFID */
1852 for (i
= 0; i
< 4; i
++) {
1853 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[i
] &=
1854 ifra
.ifra_prefixmask
.sin6_addr
.s6_addr32
[i
];
1858 if (in6_get_tmpifid(ifp
, (u_int8_t
*)randid
,
1859 (const u_int8_t
*)&ia0
->ia_addr
.sin6_addr
.s6_addr
[8], forcegen
)) {
1860 nd6log((LOG_NOTICE
, "in6_tmpifadd: failed to find a good "
1864 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[2] |=
1865 (randid
[0] & ~(ifra
.ifra_prefixmask
.sin6_addr
.s6_addr32
[2]));
1866 ifra
.ifra_addr
.sin6_addr
.s6_addr32
[3] |=
1867 (randid
[1] & ~(ifra
.ifra_prefixmask
.sin6_addr
.s6_addr32
[3]));
1870 * in6_get_tmpifid() quite likely provided a unique interface ID.
1871 * However, we may still have a chance to see collision, because
1872 * there may be a time lag between generation of the ID and generation
1873 * of the address. So, we'll do one more sanity check.
1875 for (ia
= in6_ifaddr
; ia
; ia
= ia
->ia_next
) {
1876 if (IN6_ARE_ADDR_EQUAL(&ia
->ia_addr
.sin6_addr
,
1877 &ifra
.ifra_addr
.sin6_addr
)) {
1878 if (trylimit
-- == 0) {
1880 * Give up. Something strange should have
1883 nd6log((LOG_NOTICE
, "in6_tmpifadd: failed to "
1884 "find a unique random IFID\n"));
1893 * The Valid Lifetime is the lower of the Valid Lifetime of the
1894 * public address or TEMP_VALID_LIFETIME.
1895 * The Preferred Lifetime is the lower of the Preferred Lifetime
1896 * of the public address or TEMP_PREFERRED_LIFETIME -
1899 if (ia0
->ia6_lifetime
.ia6t_vltime
!= ND6_INFINITE_LIFETIME
) {
1900 vltime0
= IFA6_IS_INVALID(ia0
) ? 0 :
1901 (ia0
->ia6_lifetime
.ia6t_vltime
-
1902 (time_second
- ia0
->ia6_updatetime
));
1903 if (vltime0
> ip6_temp_valid_lifetime
)
1904 vltime0
= ip6_temp_valid_lifetime
;
1906 vltime0
= ip6_temp_valid_lifetime
;
1907 if (ia0
->ia6_lifetime
.ia6t_pltime
!= ND6_INFINITE_LIFETIME
) {
1908 pltime0
= IFA6_IS_DEPRECATED(ia0
) ? 0 :
1909 (ia0
->ia6_lifetime
.ia6t_pltime
-
1910 (time_second
- ia0
->ia6_updatetime
));
1911 if (pltime0
> ip6_temp_preferred_lifetime
- ip6_desync_factor
){
1912 pltime0
= ip6_temp_preferred_lifetime
-
1916 pltime0
= ip6_temp_preferred_lifetime
- ip6_desync_factor
;
1917 ifra
.ifra_lifetime
.ia6t_vltime
= vltime0
;
1918 ifra
.ifra_lifetime
.ia6t_pltime
= pltime0
;
1921 * A temporary address is created only if this calculated Preferred
1922 * Lifetime is greater than REGEN_ADVANCE time units.
1924 if (ifra
.ifra_lifetime
.ia6t_pltime
<= ip6_temp_regen_advance
)
1927 /* XXX: scope zone ID? */
1929 ifra
.ifra_flags
|= (IN6_IFF_AUTOCONF
|IN6_IFF_TEMPORARY
);
1931 /* allocate ifaddr structure, link into chain, etc. */
1934 updateflags
|= IN6_IFAUPDATE_DADDELAY
;
1935 if ((error
= in6_update_ifa(ifp
, &ifra
, NULL
, updateflags
)) != 0)
1938 newia
= in6ifa_ifpwithaddr(ifp
, &ifra
.ifra_addr
.sin6_addr
);
1939 if (newia
== NULL
) { /* XXX: can it happen? */
1941 "in6_tmpifadd: ifa update succeeded, but we got "
1943 return (EINVAL
); /* XXX */
1945 newia
->ia6_ndpr
= ia0
->ia6_ndpr
;
1946 newia
->ia6_ndpr
->ndpr_refcnt
++;
1949 * A newly added address might affect the status of other addresses.
1950 * XXX: when the temporary address is generated with a new public
1951 * address, the onlink check is redundant. However, it would be safe
1952 * to do the check explicitly everywhere a new address is generated,
1953 * and, in fact, we surely need the check when we create a new
1954 * temporary address due to deprecation of an old temporary address.
1956 pfxlist_onlink_check();
1962 in6_init_prefix_ltimes(struct nd_prefix
*ndpr
)
1965 /* check if preferred lifetime > valid lifetime. RFC2462 5.5.3 (c) */
1966 if (ndpr
->ndpr_pltime
> ndpr
->ndpr_vltime
) {
1967 nd6log((LOG_INFO
, "in6_init_prefix_ltimes: preferred lifetime"
1968 "(%d) is greater than valid lifetime(%d)\n",
1969 (u_int
)ndpr
->ndpr_pltime
, (u_int
)ndpr
->ndpr_vltime
));
1972 if (ndpr
->ndpr_pltime
== ND6_INFINITE_LIFETIME
)
1973 ndpr
->ndpr_preferred
= 0;
1975 ndpr
->ndpr_preferred
= time_second
+ ndpr
->ndpr_pltime
;
1976 if (ndpr
->ndpr_vltime
== ND6_INFINITE_LIFETIME
)
1977 ndpr
->ndpr_expire
= 0;
1979 ndpr
->ndpr_expire
= time_second
+ ndpr
->ndpr_vltime
;
1985 in6_init_address_ltimes(struct nd_prefix
*new,
1986 struct in6_addrlifetime
*lt6
)
1989 /* Valid lifetime must not be updated unless explicitly specified. */
1990 /* init ia6t_expire */
1991 if (lt6
->ia6t_vltime
== ND6_INFINITE_LIFETIME
)
1992 lt6
->ia6t_expire
= 0;
1994 lt6
->ia6t_expire
= time_second
;
1995 lt6
->ia6t_expire
+= lt6
->ia6t_vltime
;
1998 /* init ia6t_preferred */
1999 if (lt6
->ia6t_pltime
== ND6_INFINITE_LIFETIME
)
2000 lt6
->ia6t_preferred
= 0;
2002 lt6
->ia6t_preferred
= time_second
;
2003 lt6
->ia6t_preferred
+= lt6
->ia6t_pltime
;
2008 * Delete all the routing table entries that use the specified gateway.
2009 * XXX: this function causes search through all entries of routing table, so
2010 * it shouldn't be called when acting as a router.
2013 rt6_flush(struct in6_addr
*gateway
, struct ifnet
*ifp
)
2015 int s
= splsoftnet();
2017 /* We'll care only link-local addresses */
2018 if (!IN6_IS_ADDR_LINKLOCAL(gateway
)) {
2023 rt_walktree(AF_INET6
, rt6_deleteroute
, (void *)gateway
);
2028 rt6_deleteroute(struct rtentry
*rt
, void *arg
)
2030 #define SIN6(s) ((struct sockaddr_in6 *)s)
2031 struct in6_addr
*gate
= (struct in6_addr
*)arg
;
2033 if (rt
->rt_gateway
== NULL
|| rt
->rt_gateway
->sa_family
!= AF_INET6
)
2036 if (!IN6_ARE_ADDR_EQUAL(gate
, &SIN6(rt
->rt_gateway
)->sin6_addr
))
2040 * Do not delete a static route.
2041 * XXX: this seems to be a bit ad-hoc. Should we consider the
2042 * 'cloned' bit instead?
2044 if ((rt
->rt_flags
& RTF_STATIC
) != 0)
2048 * We delete only host route. This means, in particular, we don't
2049 * delete default route.
2051 if ((rt
->rt_flags
& RTF_HOST
) == 0)
2054 return (rtrequest(RTM_DELETE
, rt_getkey(rt
), rt
->rt_gateway
,
2055 rt_mask(rt
), rt
->rt_flags
, 0));
2060 nd6_setdefaultiface(int ifindex
)
2064 if (ifindex
< 0 || if_indexlim
<= ifindex
)
2066 if (ifindex
!= 0 && !ifindex2ifnet
[ifindex
])
2069 if (nd6_defifindex
!= ifindex
) {
2070 nd6_defifindex
= ifindex
;
2071 if (nd6_defifindex
> 0) {
2072 nd6_defifp
= ifindex2ifnet
[nd6_defifindex
];
2077 * Our current implementation assumes one-to-one maping between
2078 * interfaces and links, so it would be natural to use the
2079 * default interface as the default link.
2081 scope6_setdefault(nd6_defifp
);